摘要
在分析传统的光斑质心亚像素细分算法误差原因的基础上,提出了一种基于"金字塔思想"分层方式的三次线形插值和最小二乘拟合的质心算法。通过多次插值对光斑图像进行逐层细分,采用最小二乘曲面拟合在每一层内进行迭代计算。结果表明,该文分层插值方法提高了图像的分辨率,减小了图像系统误差和随机噪声等因素对算法精度的影响,通过该方法获得了较高精度的图像光斑中心位置。仿真实验测试证明,分层插值拟合方法精度优于质心法和曲面拟合法等传统算法,在实际的飞行器测量实验中具有很好的实用性。
The causes for the errors of traditional centroid subpixel algorithms are theoretically analyzed and an improved centroid algorithm based on cubic convolution interpolation by pyramid algorithm and least squares fitting is presented. Several interpolations are used to subdivide the spot images and least squares surface fitting is employed to compute iteratively in each level. The image resolution is increased and the influence of system errors and random noise in image on the algorithm is reduced, and the spot center is located accurately. The simulation results show that the accuracy of the level interpolation fitting algorithm is better than centroid algorithm and surface fitting algorithm and has good practicability in aircraft measurement experiments.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2008年第5期615-618,共4页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(50275040)
关键词
质心算法
插值
最小二乘拟合
亚像素精度
centroid algorithms
interpolations
least squares fitting
subpixel precision
